PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Neural Networks, Computer”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,405 records · Page 78Linked to original sources
PubMed2006

Value at risk estimation using independent component analysis-generalized autoregressive conditional heteroscedasticity (ICA-GARCH) models.

We suggest using independent component analysis (ICA) to decompose multivariate time series into statistically independent time series. Then, we propose to use ICA-GARCH models which are computationally efficient to estimate the multivariate volatilities. The experimental results show that the ICA-GARCH models are more effective than existing methods, including DCC, PCA-GARCH, and EWMA. We also apply the proposed models to compute value at risk (VaR) for risk management applications. The backtesting and the out-of-sample tests validate the performance of ICA-GARCH models for value at risk estimation.

Computer Simulation↗
PubMed1996

Improvement of projection-type associative memories by using a two-phase recalling procedure.

In this letter, we propose a Two-Phase Recalling Procedure (TPRP) to improve the recall capability of the projection-type associative memory. It is known that the conventional projection dynamic sometimes produces spurious states that do not belong to the space spanned by the prototype vectors (memory space). The proposed TPRP provides a trapped spurious state another chance to project onto the memory space such that the recall capability of the projection-type associative memories can be greatly improved. Finally a simulation result demonstrates the effectiveness of the TPRP.

Computer Simulation↗
PubMed1993

Behaviorally based modeling and computational approaches to neuroscience.

The almost incredible advances that have recently occurred in the power of computers available to scientists in all disciplines have encouraged an explosion of neural network and behavioral models. Some of these have been constrained more by the imagination of the programmer than by rude biological facts. Their relevance for the experimental neuroscientist thus varies from case to case. Some models (e.g. Grillner's model of lamprey swimming movements) are so closely based on known neuroanatomy and neurophysiology that it becomes possible to generate and test precise experimental predictions. Other models (such as MURPHY and NOMAD) use neurobiological principles in their architectures, but do not portray any particular organism. Although it is harder to relate the study of these models of the study of real animals, they fulfill an important explanatory role. They make possible insights into how behavior is controlled by neuronal activity that would be unobtainable in real animals using present methods. Thus, even the excesses of neural modeling have provided a useful impetus to what is undoubtedly a most promising approach to integrating data from the various disciplines concerned with behavior and the mind. The problems have been pointed out by many authors (see citations in our introduction), and a phase of more critical evaluation appears to have begun. We hope that our brief survey of models based on widely different theoretical approaches, but all aimed at explaining behavior, will encourage critical comparisons to be made. As in more mature fields, such as thermodynamics, we can expect that more complete models will force an evaluation of theoretical hypotheses against the entire body of available evidence, rather than just a few pertinent test cases. Such evaluation will make possible a much more rigorous exclusion of invalid or inconsistent theoretical ideas. From such studies, a much smaller, but more robust, set of basic principles can be expected to emerge. From the perspective afforded by our own modeling studies, it appears essential that modeling be informed by a general theory of brain function. In this work, the theory of neuronal group selection provides a useful basis for further work by virtue of its consistency with basic evolutionary and physiological principles and the power of the selection paradigm to shape neural networks in behaviorally adaptive directions.

Animals↗
PubMedDate not supplied

PAPNET for cervical cytology screening. Experience in Greece.

OBJECTIVE: To validate the efficacy of PAPNET for cervical cytology screening. STUDY DESIGN: From a total of 50,000 gynecologic cases we selected 24 primarily negative smears from patients who, during follow-up, developed a precancerous lesion or cancer in a short time, ranging from a few months to two years. These negative smears were rescreened by the PAPNET automated system and reevaluated by two observers. A third observer blindly reevaluated the same cases manually. RESULTS: Reevaluation after PAPNET screening showed one false negative smear, for a rate comparable to that of manual reevaluation, which also reclassified the same case. The time required to interpret a smear using the PAPNET system was approximately one minute, while the time required for conventional manual screening was approximately five. CONCLUSION: PAPNET is an efficient and fast screener for the detection of precancerous cervical lesions and cancer.

False Negative Reactions↗
PubMedDate not supplied

Simulation of primary cervical cancer screening by the PAPNET system in an unscreened, high-risk community.

OBJECTIVE: To evaluate the performance of the PAPNET system as a primary cervical cancer screening modality in an unscreened population with a high prevalence of cervical cancer and its precursor lesions. STUDY DESIGN: Consecutive cervical smears from 3,106 women, screened and reported in the usual manner, were submitted for analysis by the PAPNET system. The original manual screening diagnoses were compared with those obtained by PAPNET analysis. By inclusion of normal and abnormal smears, this evaluation not only provided quality assurance for the laboratory but also simulated primary screening by automation. RESULTS: Comparison of the two methods of screening showed statistically significant superiority of the PAPNET over conventional screening (89.6% vs. 63.8%, respectively) in low grade lesions, including atypical squamous and atypical glandular cells of uncertain significance (ASCUS and AGUS, respectively) and low grade squamous intraepithelial lesion. Conversely, there was no significant difference between PAPNET and manual detection (87.5% vs. 94.6%) for more significant abnormalities, including high grade squamous intraepithelial lesions and invasive carcinoma. CONCLUSION: The PAPNET system, which would probably not be affordable as a quality assurance modality only in the public health sector of this country, was shown to be more than sufficiently effective as a primary screening method for the large numbers of women likely to undergo cervical cancer screening in anticipated mass population programs.

Automation↗
PubMedDate not supplied

Clinical validation of interactive cytologic screening. Automating the search, not the interpretation.

OBJECTIVE: To evaluate the effectiveness of the PAPNET Testing System in improving cervical smear screening accuracy as measured during routine screening. STUDY DESIGN: A review of several recently published studies on the effectiveness of PAPNET testing in clinical use. System performance was retrospectively and prospectively evaluated and compared to the conventional method of testing. The use of PAPNET rescreening to identify smears having more significant abnormalities than had been initially diagnosed was investigated. RESULTS: New studies demonstrate that PAPNET testing increases the detection of significant lesions and improves cytotechnologist performance in routine clinical practice. Less variable diagnostic results are achieved with PAPNET testing as compared to conventional methods. The system is effective in identifying atypical smears having more serious abnormalities. CONCLUSION: When used as a supplemental method in routine practice, as studies have shown, PAPNET testing can reduce interobserver variability and increase the detection of abnormality.

Automation↗
PubMed2005

Interaction of face and voice areas during speaker recognition.

Face and voice processing contribute to person recognition, but it remains unclear how the segregated specialized cortical modules interact. Using functional neuroimaging, we observed cross-modal responses to voices of familiar persons in the fusiform face area, as localized separately using visual stimuli. Voices of familiar persons only activated the face area during a task that emphasized speaker recognition over recognition of verbal content. Analyses of functional connectivity between cortical territories show that the fusiform face region is coupled with the superior temporal sulcus voice region during familiar speaker recognition, but not with any of the other cortical regions normally active in person recognition or in other tasks involving voices. These findings are relevant for models of the cognitive processes and neural circuitry involved in speaker recognition. They reveal that in the context of speaker recognition, the assessment of person familiarity does not necessarily engage supramodal cortical substrates but can result from the direct sharing of information between auditory voice and visual face regions.

Acoustic Stimulation↗
PubMed2000

Estimating functions of independent component analysis for temporally correlated signals.

This article studies a general theory of estimating functions of independent component analysis when the independent source signals are temporarily correlated. Estimating functions are used for deriving both batch and on-line learning algorithms, and they are applicable to blind cases where spatial and temporal probability structures of the sources are unknown. Most algorithms proposed so far can be analyzed in the framework of estimating functions. An admissible class of estimating functions is derived, and related efficient on-line learning algorithms are introduced. We analyze dynamical stability and statistical efficiency of these algorithms. Different from the independently and identically distributed case, the algorithms work even when only the second-order moments are used. The method of simultaneous diagonalization of cross-covariance matrices is also studied from the point of view of estimating functions.

Algorithms↗
PubMed2000

Nonholonomic orthogonal learning algorithms for blind source separation.

Independent component analysis or blind source separation extracts independent signals from their linear mixtures without assuming prior knowledge of their mixing coefficients. It is known that the independent signals in the observed mixtures can be successfully extracted except for their order and scales. In order to resolve the indeterminacy of scales, most learning algorithms impose some constraints on the magnitudes of the recovered signals. However, when the source signals are nonstationary and their average magnitudes change rapidly, the constraints force a rapid change in the magnitude of the separating matrix. This is the case with most applications (e.g., speech sounds, electroencephalogram signals). It is known that this causes numerical instability in some cases. In order to resolve this difficulty, this article introduces new nonholonomic constraints in the learning algorithm. This is motivated by the geometrical consideration that the directions of change in the separating matrix should be orthogonal to the equivalence class of separating matrices due to the scaling indeterminacy. These constraints are proved to be nonholonomic, so that the proposed algorithm is able to adapt to rapid or intermittent changes in the magnitudes of the source signals. The proposed algorithm works well even when the number of the sources is overestimated, whereas the existent algorithms do not (assuming the sensor noise is negligibly small), because they amplify the null components not included in the sources. Computer simulations confirm this desirable property.

Algorithms↗
PubMed2003

Parameter estimation of sigmoid superpositions: dynamical system approach.

Superposition of sigmoid function over a finite time interval is shown to be equivalent to the linear combination of the solutions of a linearly parameterized system of logistic differential equations. Due to the linearity with respect to the parameters of the system, it is possible to design an effective procedure for parameter adjustment. Stability properties of this procedure are analyzed.

Computer Simulation↗
PubMed2003

Visual development and the acquisition of motion velocity sensitivities.

We consider the hypothesis that systems learning aspects of visual perception may benefit from the use of suitably designed developmental progressions during training. Four models were trained to estimate motion velocities in sequences of visual images. Three of the models were developmental models in the sense that the nature of their visual input changed during the course of training. These models received a relatively impoverished visual input early in training, and the quality of this input improved as training progressed. One model used a coarse-to-multiscale developmental progression (it received coarse-scale motion features early in training and finer-scale features were added to its input as training progressed), another model used a fine-to-multiscale progression, and the third model used a random progression. The final model was nondevelopmental in the sense that the nature of its input remained the same throughout the training period. The simulation results show that the coarse-to-multiscale model performed best. Hypotheses are offered to account for this model's superior performance, and simulation results evaluating these hypotheses are reported. We conclude that suitably designed developmental sequences can be useful to systems learning to estimate motion velocities. The idea that visual development can aid visual learning is a viable hypothesis in need of further study.

Artifacts↗
PubMed2004

Rapid processing and unsupervised learning in a model of the cortical macrocolumn.

We study a model of the cortical macrocolumn consisting of a collection of inhibitorily coupled minicolumns. The proposed system overcomes several severe deficits of systems based on single neurons as cerebral functional units, notably limited robustness to damage and unrealistically large computation time. Motivated by neuroanatomical and neurophysiological findings, the utilized dynamics is based on a simple model of a spiking neuron with refractory period, fixed random excitatory interconnection within minicolumns, and instantaneous inhibition within one macrocolumn. A stability analysis of the system's dynamical equations shows that minicolumns can act as monolithic functional units for purposes of critical, fast decisions and learning. Oscillating inhibition (in the gamma frequency range) leads to a phase-coupled population rate code and high sensitivity to small imbalances in minicolumn inputs. Minicolumns are shown to be able to organize their collective inputs without supervision by Hebbian plasticity into selective receptive field shapes, thereby becoming classifiers for input patterns. Using the bars test, we critically compare our system's performance with that of others and demonstrate its ability for distributed neural coding.

Afferent Pathways↗
PubMed1997

Detecting synchronous cell assemblies with limited data and overlapping assemblies.

Two statistical methods-cross-correlation (Moore et al. 1966) and gravity clustering (Gerstein et al. 1985)-were evaluated for their ability to detect synchronous cell assemblies from simulated spike train data. The two methods were first analyzed for their temporal sensitivity to synchronous cell assemblies. The presented approach places a lower bound on the amount of data required to detect a synchronous assembly. On average, both methods required the same minimum amount of recording time to detect significant pairwise correlations, but the gravity method exhibited less variance in the recording time. The precise length of recording depends on the consistency with which a neuron fires synchronously with the assembly but was independent of the assembly firing rate. Next, the statistical methods were tested with respect to their ability to differentiate two distinct assemblies that overlapped in time and space. Both statistics could adequately differentiate two overlapping synchronous assemblies. For cross-correlation, this ability deteriorates quickly when considering three or more simultaneously active, overlapping assemblies, whereas the gravity method should be more flexible in this regard. The work demonstrates the difficulty of detecting assembly phenomena from simultaneous neuronal recordings. Other statistical methods and the detection of other types of assemblies are also discussed.

Action Potentials↗
PubMed2005

Monosynaptic connections between pairs of spiny stellate cells in layer 4 and pyramidal cells in layer 5A indicate that lemniscal and paralemniscal afferent pathways converge in the infragranular somatosensory cortex.

Monosynaptic interlaminar connections between spiny stellate cells in layer 4 (L4), the main cortical recipient layer for thalamic projections, and pyramidal cells in layer 5A (L5A), one of the main cortical output layers, were examined anatomically and functionally by paired recordings in acute brain slices. The somata of pairs forming interlaminar L4-to-L5A connections were located predominantly close to or directly under the barrel-septum wall in layer 4. Superposition of spiny stellate axon arbors and L5A pyramidal cell dendritic arbors suggested an innervation domain underneath an L4 barrel wall. Functionally, the L4-to-L5A connections were of high reliability and relatively low efficacy, with a unitary EPSP amplitude of 0.6 mV, and the connectivity was moderately high (one in seven pairs tested was connected). The EPSP amplitude was weakly depressing (paired-pulse ratio of approximately 0.8) during repetitive presynaptic action potentials at 10 Hz. The existence of monosynaptic L4-to-L5A connections indicates that the specific "lemniscal" thalamic input from the ventro-basal nucleus of the thalamus to the cortex and the more unspecific "paralemniscal" afferent thalamic projections from the posterior medial nucleus of the thalamus merge already at an initial stage of cortical signal processing. These monosynaptic connections establish a monosynaptic coupling of the input to the cortex and its output, thereby effectively bypassing the supragranular layers.

Afferent Pathways↗
PubMed2006

Integrated neural processes for defining potential actions and deciding between them: a computational model.

To successfully accomplish a behavioral goal such as reaching for an object, an animal must solve two related problems: to decide which object to reach and to plan the specific parameters of the movement. Traditionally, these two problems have been viewed as separate, and theories of decision making and motor planning have been developed primarily independently. However, neural data suggests that these processes involve the same brain regions and are performed in an integrated manner. Here, a computational model is described that addresses both the question of how different potential actions are specified and how the brain decides between them. In the model, multiple potential actions are simultaneously represented as continuous regions of activity within populations of cells in frontoparietal cortex. These representations engage in a competition for overt execution that is biased by modulatory influences from prefrontal cortex. The model neural populations exhibit activity patterns that correlate with both the spatial metrics of potential actions and their associated decision variables, in a manner similar to activities in parietal, prefrontal, and premotor cortex. The model therefore suggests an explanation for neural data that have been hard to account for in terms of serial theories that propose that decision making occurs before action planning. In addition to simulating the activity of individual neurons during decision tasks, the model also reproduces key aspects of the spatial and temporal statistics of human choices and makes a number of testable predictions.

Computational Biology↗
PubMed2006

Application of an ecological framework linking scales based on self-thinning.

Barnes and Roderick developed a generic, theoretical framework for vegetation modeling across scales. Inclusion of a self-thinning mechanism connects the individual to the larger-scale population and, being based on the conservation of mass, all mass flux processes are integral to the formulation. Significantly, disturbance (both regular and stochastic) and its impact at larger scales are included in the formulation. The purpose of this paper is to illustrate how this model can be used to predict patch and ecosystem dry mass, and consequently system carbon. Examples from pine plantations and mixed forests are considered, with these applications requiring estimates of system carrying capacity and the growth rates of individual plants. The results indicate that the model is relatively simple and straightforward to apply, and its predictions compare well with the data. A significant feature of this approach is that the impact of local scale data on the dynamics of larger patch and ecosystem scales can be determined explicitly, as we show by example. Further, the general formulation has an analytic solution based on characteristics of the individual, facilitating practical and predictive application.

Adaptation, Physiological↗
PubMed2000

Odour recognition and segmentation by a model olfactory bulb and cortex.

We present a model of an olfactory system that performs odour segmentation. Based on the anatomy and physiology of natural olfactory systems, it consists of a pair of coupled modules, bulb and cortex. The bulb encodes the odour inputs as oscillating patterns. The cortex functions as an associative memory: when the input from the bulb matches a pattern stored in the connections between its units, the cortical units resonate in an oscillatory pattern characteristic of that odour. Further circuitry transforms this oscillatory signal to a slowly varying feedback to the bulb. This feedback implements olfactory segmentation by suppressing the bulbar response to the pre-existing odour, thereby allowing subsequent odours to be singled out for recognition.

Afferent Pathways↗